1. Field of the Invention
This invention relates to shutting down library managers and more particularly relates to shutting down one or more library managers while maintaining a library system online.
2. Description of the Related Art
Currently, updating code in a hierarchical data storage system such as the International Business Machine (“IBM”) 3494 Tape Library Dataserver is a disruptive process. FIG. 1 illustrates a typical library system 100. The system 100 may include one or more hosts 102, a device controller 104 in communication with a host 102, one or more library managers 106 in communication with the device controller 104, and one or more storage devices 108 in communication with the device controller 104, and the library managers 106.
A dual library manager system may be termed a Highly Available (“HA”) library manager system where a first library manager 106a is a primary and a second library manager 106b is a backup that can quickly be brought online if the first library manager 106a fails. A library system 100 may include also a single library manager 106a and/or a single storage device 108. Alternatively, the system 100 includes an array of device controllers 104, library managers 106 and storage devices 108. A storage device 108 is typically a computer readable tape, a compact disk, a removable disk drive, or other storage device managed by a library manager 106. One skilled in the art will recognize other applicable library systems 100.
Currently, object code (also referred to as code) implementing the library manager 106 is typically not running when new code is activated or for other maintenance operations. To update the library manager code, the library manager 106 typically is taken offline and then shutdown. As used herein, offline refers to a state where the library system 100 is not accessible by a host 102. As used herein, shutdown refers to a state where code is not in a condition to be executed and the code and other device components may be modified, updated, or replaced. In a dual library manager system 100, the code can be updated or minor maintenance operations performed on a first offline library manager 106a. Then, the updated first library manager 106a can be brought online while a second library manager 106b is taken offline in a switchover process. This can be accomplished while storage devices 108 remain online in relation to the host 102.
Unfortunately, this switchover process is limited. Specifically the conventional switchover process does not allow for code updates of a single library manager system or for a dual library manager system 100 when both library managers need to be varied offline at the same time for a major code update or maintenance operation.
For single library manager systems or for systems with more than one library manager 100 where a code update requires all library managers to be varied offline at the same time, currently the whole library system 100 is taken offline with respect to a host 102. Taking the library system 100 offline can be a very time consuming process due to a possible long completion time of data transaction commands and library manager commands currently queued. While the library system 100 is being varied offline, the library system 100 is unavailable for new data transactions or library manager transactions. Scheduling several hours for a busy library system 100, such as an HA library system 100, to be offline may be a difficult task and may cause costly delays while the system 100 is unavailable.
From the foregoing discussion, it should be apparent that a need exists for an apparatus, system, and method to shutdown library managers while maintaining the library system in a partially online state. Beneficially, such an apparatus, system, and method would greatly reduce the time required for a maintenance operation or code update, would allow major code updates or maintenance operations while the library system remains in a partially online state, and would maintain a host in a partially online state allowing data operations with previously mounted storage devices to continue. Also, after major code updates or maintenance operations are complete, the library is brought fully online and all normal operations resume.